From my viewpoint there are not many truly big changes to Scrum or Agile practice at all. There are lots of smaller changes that change how we do things, but not what we are trying to do. I'll come up with some examples at random:
- Scrum as practiced adopted some XP ideas, including user stories, tasks, estimation, and release planning.
- The XP community, at least as represented by Ron and Chet, no longer favor tasks. Nor do we favor story estimation. We are opposed to release planning, though it cannot always be avoided. (The reason is that release planning militates against what works best in Agile, namely steering the project to success by judicious choice of features to build.)
- Technical practices, again borrowed from XP, are beginning to be understood to be essentially necessary for successful Scrum projects. The Scrum Alliance Certified Scrum Developer program reflects this understanding.
- The Product Owner job is harder than anyone imagined.
- The backlog grooming process is more important than most people recognize at first.
- Lean techniques, especially kanban techniques, can be quite valuable in improving the process.
- There are situations where the project may be better served by continuous flow rather than using iterations.
I'm sure there are many more. Note that none of these have substantial impact at the "policy" level. They are just recognitions of better ways of doing things.
There are also some important understandings about what goes wrong.
- Lack of an empowered business-focused Product Owner is deadly to project success.
- Scrum Master is not a management or project management position.
- Pressure to deliver "more" is almost instantly damaging to project quality, and inevitably means the project will deliver fewer features, and more defects.
- It is not very productive to divide work across technical boundaries, with Scrum teams working on bits of infrastructure or integration or some other technical breakdown.
- Failure to follow good technical practice results in slower delivery of inferior software.
Again, there are many more ways to go wrong.
It turns out that, in my opinion, addressing these concerns do two things. First, you move more in the direction of the Agile Manifesto.Second, and far more important, you move your project toward more likely success.
So ... in the opinion of the Authors, the Manifesto stands up pretty well. There has been much learning about how to do things.